$oprano

$oprano

Local
Статус
Offline
Joined
6/24/21
Messages
38
Reaction score
1
Have you long dreamed of becoming your mother’s hacker so you can undress your girlfriends using neural networks? - then the article is just for you

To start undressing your girlfriends, you first need to download and install the neural network. There are two ways to do this: Portable (on your computer) and GoogleColab (using the power of Google). The portable version is more stable and easier to work with, but you will need a computer that can handle at least the third Witcher at high speeds.

Installation portable

1. Follow the link click (https://github.com/serpotapov/stable-diffusion-portable) VT - click (https://www.virustotal.com/gui/file...68b4aff9872b4ba120a3dc6a1682b972c58db6cd65975) click on the green “code” button, then “download ZIP” and wait for the archive to download. Unpack the downloaded archive to any disk.

2. Next, go to the unpacked archive and click on “webui-user-first-run”, the installation will start and the program will automatically start downloading all the necessary components. You will have to wait a long time (20-40 minutes, especially if the Internet is weak and the computer is not very powerful. Sometimes errors may appear, do not pay attention to them.

3. Next you need to download the models, for this we go here: click (to download (https://civitai.com/models/2661?modelVersionId=15670) you will need to log in) and upload them to the Stable-diffusion folder along the path stable-diffusion-portable-main>models>Stable-diffusion

UPD: if the models do not suit you and you still can’t do something normal, try looking for other models on the same site, there are many different ones, I think by trial and error you will find something for yourself

4. After all the preparations, the neural network can be launched; this is done using the webui-user.bat file (it will appear in the stable-diffusion-portable-main folder). First, the console will launch, and then a tab in the browser, it is through it that we will generate deepfakes! Important: closing the console = closing the program

COLAB version

1. Colab installation

For this method we need an account on Google Drive. Follow this link click (https://colab.research.google.com/g...ain/fast_stable_diffusion_AUTOMATIC1111.ipynb) and click the "Copy to Disk" button ". Now we will work in the "Copy of notepad..." tab Screen (https://lztcdn.com/files/6ca92a3e02e4cc7ee1b6e5d5a4e420a4f6af01854dc6c8427ba2ecc3abb3140f.webp)

2. Colab installation

We launch the first three processes in turn (connect, install, requiements) After they are completed, check whether new folders have appeared on your disk - if they have appeared, everything worked as it should

3. Colab installation

Now you need to download the model from this link (https://civitai.com/models/2661?modelVersionId=15670) and upload it to Google Drive. After that, copy the path to the model as shown here (https://lztcdn.com/files/f22b3450096391b9f7899c9c81c6637f20b949c16d190645f8894608f881d736.webp) and paste this path into the next process and run it

4. The launch of the colab version itself

We launch the ControlNet and Start Stable-Diffusion process, it can load in a couple of minutes. After a while, the process will provide a link through which we can work with the neural network for 72 hours. After three days, the Start Stable-Diffusion process will have to be started again to get a new link!

Setting up and working with a neural network

It doesn’t matter if you have a portable version or a Google Colab version, the settings will be the same.
We display everything as on the screen (https://lztcdn.com/files/3b64505ea3a10e5a88615f09549ca645c4c55590132b5eea5038ebbdcbbd8d59.webp)

Select the img2img tab.
In it we select Inpaint

We change only these settings: Sampling method to DPM++ SDE, Sampling steps set in the range 20-25, Denoising strength 75-90 (this setting determines how much the image will change), the expansion of the generated image is set to no more than 1280.

And don't forget to paint over the place we want to change

Negatives are what the neural network will not draw: (deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, (mutated hands and fingers:1.4), disconnected limbs, mutation, mutated, ugly, disgusting, blurry, amputation

Prompts are what the neural network will have as a priority, the closer it is to the beginning, the more important it will be for the neural network, usually naked is enough, if the girl is thin, then we write skinny, you can play with it and sooner or later you will figure it out

After you have entered prom and negatives, configured the sliders as you need, press the Generate button and enjoy, if the result does not suit you, try tweaking the settings, changing the Sampling method or Masked content

Enjoying the photo
 
Top